Sparse Representation Face Recognition Algorithm Based On Image Blocks

Face recognition is considered as an important technology of modern biological information recognition. Most of the exsiting face recognition methods need complex image processing and feature extraction, the choice of features has a great effect on the recognition rate, and it lacks robust about occlusion and corruption. These problems often make the existing recognition methods restricted in the application.With advantages like high recognition rate and strong robustness, more and more researchers take attention on sparse represention. Sparse represention is the important theory of compressive sensing, increase the efficiency of compression. Sparse represention has a unique advantage in classification, it makes feature selection is no longer critical. The work of this paper is as follows:(1) We propose a novel framework of robust face recognition based on the sparse representation. Firstly, image is divided into modules and each module is processed separately to determine its reliability. We propose to use the modular sparsity and residual jointly to determine the modular reliability. Secondly, a reconstructed image from the modules weighted by their reliability is formed for the robust recognition. Compared with the traditional algorithm, the proposed recognition algorithm is improved efficiently.(2) Human face with varying expression and occlusion, as well as disguise and illumination problem is the major problem affecting human face recognition rate. Through the analysis of the above problems, the paper proposes a novel algorithm of face recognition based on local structural sparse representation. This algorithm adopts a novel low-rank matrix method with structural incoherence and uses DCT method jointly deal with the problems of occlusion, disguise and illumination variations in face image. After that, the algorithm adopts a unique overlapping partition method to the processed image, using the redundant information to improve the recognition rate of the algorithm. In the classification phase, the algorithm effectively improves the recognition speed by using the idea of alignment pooling. The algorithm has been done many experiments on the standard face databases. Compared with the related state-of-the-art methods, the experiment accuracy and efficiency results verify the advancement of the proposed method.(3) Face recognition based on sparse representation (SRC) is mainly for single input image, the face image taken by individuals under different conditions can only be one by one through detection identification system to determine the identity. However, when the dimension of training set is very high, the recognition speed may slow down greatly. So in this paper, we propose a face recognition approach based on image reconstruction and hashing. The image reconstruction based on sparse coding improves upon the sparse representation classification framework (SRC) and extends the current SR (Sparse Coding) framework to classification problems with multiple input samples. The image reconstruction essentially looks for a more representation image from the multiple input samples. Simultaneously, we propose a novel face recognition approach based on hashing and OMP (Orthogonal Matching Pursuit). Since hashing can preserve the restrictive isometry property and OMP can improve the sparsity of the coefficients,the proposed work improves the recognition rate and recognition speed We conduct extensive experiments on publicly available face databases and the experimental results demonstrate the accuracy and efficiency of the proposed approach.